yawfrempong@google.com chrome-counter-abuse-alerts@google.com Records the outcome of a download on Android, with respect to Safe Browsing download protection, up until the point of sending a ClientDownloadRequest. This is logged exactly once per download for any download on Android, regardless of Safe Browsing state. This does not record the outcome of the ClientDownloadRequest itself, or the actual verdict if a ClientDownloadResponse was received. yawfrempong@google.com chrome-counter-abuse-alerts@google.com Records the result of getting the referring app info for a download protection ClientDownloadRequest sent on Android. This is recorded after the request is finalized but before it is sent. drubery@chromium.org chrome-counter-abuse-alerts@google.com Records the extension of files downloaded with auto-open {State}. This can be used to decide the impact of changing the `auto_open_hint` in the file type policy. It is logged on every successful download. yawfrempong@google.com chrome-counter-abuse-alerts@google.com Records the events related to ephemeral warning cancellation. Compare the bucket count between CancellationScheduled and CancellationTriggered to see how often the delayed task actually got triggered after one hour. The volume of the rest of the buckets should sum up to the CancellationTriggered bucket. vakh@chromium.org chrome-counter-abuse-alerts@google.com mattm@chromium.org Records the reason why downloads are marked as being malicious or clean by the improved SafeBrowsing binary download protection. Note that UNSUPPORTED_URL_SCHEME was split out of the INVALID_URL bucket in M41. The NOT_BINARY_FILE check was moved before the UNSUPPORTED_URL_SCHEME check in M42. The ALLOWLISTED_URL check was moved to before NOT_BINARY_FILE in M62. This is logged after download scanned by Safe Browsing. This metric is used to populate a dashboard on go/crsb-site. yawfrempong@google.com chrome-counter-abuse-alerts@google.com Records the reason each file was uploaded to Safe Browsing for deep scanning. This is logged on every deep scan. domfc@chromium.org chrome-counter-abuse-alerts@google.com Records the type of scan each time a download item starts a deep scan. This is logged on every deep scan. vakh@chromium.org chrome-counter-abuse-alerts@google.com mattm@chromium.org Records a histogram of how often users download a file with a file extension that is possibly dangerous (e.g., exe, class). yawfrempong@google.com chrome-counter-abuse-alerts@google.com Records the total time it takes for the SafeBrowsing download service to check whether the content of a download which underwent {Analysis} analysis is malicious or not, including file feature extraction, allowlist checking, and server ping. This histogram only includes checks that sent a ping to the SafeBrowsing server. It does not include requests that were cancelled, but does include requests that received a bad response. drubery@chromium.org chrome-counter-abuse-alerts@google.com Records the time it takes for the SafeBrowsing download service ping. It is not recorded for requests that were cancelled. Warning: this histogram was expired from 2021-05-06 to 2021-12-02; data may be missing. yawfrempong@google.com chrome-counter-abuse-alerts@google.com Records the net error or HTTP response code after sending a CheckClientDownloadRequest. yawfrempong@google.com chrome-counter-abuse-alerts@google.com Records the time it takes for the SafeBrowsing download service to check the safety of the contents of a download which underwent analysis for {Analysis}. This only includes file feature extraction. Also see SBClientDownload.DownloadRequestDuration. yawfrempong@google.com chrome-counter-abuse-alerts@google.com Records events related to user interaction with local decryption prompts. This includes when the prompt is shown, when it is accepted or rejected, and when the provided password is incorrect. domfc@chromium.org chrome-counter-abuse-alerts@google.com Records the number of files corresponding to a single save package being scanned. This is logged on every save package deep scan. liu@chromium.org chrome-counter-abuse-alerts@google.com Records how often users in a population consented to deep scanning are actually prompted for deep scanning of {Encryption}. This is logged once per download by an eligible user. liu@chromium.org chrome-counter-abuse-alerts@google.com Records how often users in a population consented to immediate deep scans are actually prompted for this feature. This is logged once per download by an eligible user. liu@chromium.org chrome-counter-abuse-alerts@google.com Records how often users in a population consented to local decryption are actually prompted for this feature. This is logged once per download by an eligible user. xinghuilu@chromium.org yawfrempong@google.com chrome-counter-abuse-alerts@google.com Records the tailored warning type that is shown. Logged when a download warning was shown in the shelf/bubble. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records the attributes of a download that is allowed on user gesture. Logged each time a download happens and the danger level is ALLOW_ON_USER_GESTURE. To get the percentage of a specific attribute, divide the number of record of that attribute by the number of record of TOTAL_TYPE_CHECKED. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records whether the download was triggered by user gesture when a download warning for {DangerType} is {Action}. The bypassed metrics are recorded when the warning is either bypassed from the download shelf or from the downloads page. By combining the shown and bypassed metrics for a specific danger type, we can calculate whether the bypass rate is different between user-initiated and automatically triggered downloads. This metric is used to populate a dashboard on go/crca-site. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records whether the download URL is HTTPS when a download warning for {DangerType} is {Action}. The bypassed metrics are recorded when the warning is either bypassed from the download shelf or from the downloads page. By combining the shown and bypassed metrics for a specific danger type, we can calculate whether the bypass rate is different between HTTPS downloads and non-HTTPS downloads. This metric is used to populate a dashboard on go/crca-site. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records the type of the file when a download warning for {DangerType} is {Action}. The bypassed metrics are recorded when the warning is either bypassed from the download shelf or from the downloads page. By combining the shown and bypassed metrics for a specific danger type, we can calculate the bypass rate of different file types. This metric is used to populate a dashboard on go/crca-site. drubery@chromium.org chrome-counter-abuse-alerts@google.com Records a more granular reason why we failed to unpack a {ArchiveType} file. This is recorded on every download of an {ArchiveType} file. drubery@chromium.org chrome-counter-abuse-alerts@google.com Records events related to user interaction with deep scanning prompts. This is recorded a few times during the process of each prompt for deep scanning: - Once when the prompt is shown. - Once when the prompt is accepted/bypassed. - Once when the scan is completed, either through cancelation or completion. This is logged for consumer scans of {Encryption}. Note: Logging of PromptAccepted was lost in M118. "Scan canceled" and "Scan deleted" were added in M121. "Prompt accepted" was split into "Prompt accepted" and "Prompt accepted from chrome://downloads" in M121. yawfrempong@google.com chrome-counter-abuse-alerts@google.com Records the result of deep scanning {Encryption}, split by the reason the file was uploaded for scanning. This is logged after every deep scan of a download. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether adding phishing report times to profile preferences list before sending ClientPhishingRequest is successful or not. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the duration of running the TfLite classification, which is one step in applying the TfLite model for phishing detection. This model is applied roughly once per navigation. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the ongoing request type that blocked a new request of type {RequestType} due to the new request having a lower or same priority tier. Logged in ClientSideDetectionHost::MaybeStartPreClassification when a new trigger occurs while another classification is already in progress, whether in preclassification check or renderer classification. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com When a renderer-side classification returns the error "CLASSIFIER_NOT_READY", this histogram is logged with whether or not the browser process had a valid model. This can help to identify problems distributing the model from browser process to the renderers. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com The counts for various reasons why an in-progress phishing classification was canceled. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com The counts for various reasons why an in-progress phishing image embedding was canceled. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records events in the phishing classifier including reasons that prevents phishing detection to complete on the renderer side. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the success rate of renderer-side phishing classification after retries when the model is not available. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the result of classification in visual TfLite model thresholds comparison on the browser side xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records the result of async check triggering CSD-Phishing classification forced by the request from real-time URL lookup. Recorded each time async Safe Browsing check is completed (note: this only happens when the check goes through async check tracker, which only happens if the Safe Browsing check is slow). nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the request type of an in-progress client-side detection process that was interrupted by a new page navigation. The process is considered in-progress from the completion of pre-classification until just prior to sending a network request or the process is otherwise completed/cancelled. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the event in Client Side Detection Host class. Record each time the flow hits the event listed in ClientSideDetectionEvent. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records client_side_detection_type in the verdict which denotes which process is responsible for sending the ping to the server. This histogram is logged once on every successful classification. liu@chromium.org chrome-counter-abuse-alerts@google.com Records the volume of analyzed clipboard payloads and whether they're deemed likely to be malicious. liu@chromium.org chrome-counter-abuse-alerts@google.com Records the duration it took to process the clipboard payload and analyze its suspicion levels. This is recorded each time the clipboard payload is processed for client-side detection. liu@chromium.org chrome-counter-abuse-alerts@google.com Records the duration it took to split the clipboard payload string into tokens. This is recorded each time the clipboard payload is processed for client-side detection. liu@chromium.org chrome-counter-abuse-alerts@google.com Records the number of suspicious tokens discovered in the clipboard payload. This is recorded each time the clipboard payload is processed for client-side detection. liu@chromium.org chrome-counter-abuse-alerts@google.com Records the number of tokens created when splitting the clipboard payload. This is recorded each time the clipboard payload is processed for client-side detection. liu@chromium.org chrome-counter-abuse-alerts@google.com Records the number of URLs extracted from a suspicious clipboard payload. This is recorded each time the clipboard payload is processed for client-side detection. liu@chromium.org chrome-counter-abuse-alerts@google.com Records the length of the clipboard payload in characters when a clipboard copy API is triggered on the page. nwokedi@chromium.org bcb@chromium.org chrome-counter-abuse-alerts@google.com Records a credit card form event that may trigger a CSD ping, with an enum that represents the permutation of details for that event. This is recorded on every event that may trigger a CREDIT_CARD_FORM classification attempt, after checking for ESB, and after filtering to only the first event of a URL visit. {EventName} nwokedi@chromium.org bcb@chromium.org chrome-counter-abuse-alerts@google.com Records a credit card form event that may trigger a CSD ping, with an enum that represents the permutation of details for that event. This is recorded on every event that may trigger a CREDIT_CARD_FORM classification attempt, after checking for ESB. {EventName} nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the times we could have used the CSD images cache to reuse the images for CSD-Phishing classification. drubery@chromium.org chrome-counter-abuse-alerts@google.com Records whether the shared memory region created to hold the flatbuffer was created successfully on not. This is logged on dynamic model update (which is rare), or on each startup. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the status when we create a FlatBuffer scorer object for the client-side phishing detection classifier. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Log whether CSD-Phishing pings contain the visual features image. This is only logged when the ping should contain the image (i.e. when SBClientPhishing.VisualFeaturesClearReason is kCanExtractVisualFeatures) and after it's determined that the ping is eligible to be sent. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the duration it takes for high confidence allowlist check to complete. This occurs in the client side detection PreClassificationCheck. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Logs whether the URL was on the high-confidence allowlist. Logged when the post classification server model verdict comes back to be phishy. nwokedi@chromium.org bcb@chromium.org chrome-counter-abuse-alerts@google.com Records the duration it takes for calling HistoryService.GetDailyVisitsToOrigin. This occurs for some ping types when determining whether to proceed to preclassification. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records events in the phishing image embedder including reasons that prevents image embedding process to complete on the renderer side. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether the loaded page's image embedding was captured. This is logged after any trigger models have finished classification and the client side detection type is `IMAGE_EMBEDDING_MATCH`. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the size of the image embedding list but displays the bucket the size entered which may be different than the size. This is logged when a new image embedding model is fetched, or on each startup. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the version of the image embedding list. This is logged when a new image embedding model is fetched, or on each startup. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records how long it takes the Flatbuffer scorer to load the TfLite model file from the memory mapped file into a string. This blocks the main thread, so we need to ensure it is small. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Logs whether the image embedding model version is matched or not so that the correctly paired trigger and image embedding model are sent together to the renderer process. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether the intelligent scan has a successful response. This is logged in the callback after the intelligent scan call in the host class that was requested by {IntelligentScanRequestType}. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Logs the size (in bytes) of the inner text extracted from the page during the intelligent scan. We can use this histogram to inform the size of the context window we need. Note that to convert bytes to tokens, we need to estimate the number of letters in a word and the number of tokens in a word. This is logged when the intelligent scan is inquired and requested by {IntelligentScanRequestType}. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether or not the intelligent scan is ongoing when a new preclassification happens. This is logged when a new request happens through a new preclassification check and the current existing scan will be destroyed. The {RequestType} will be the triggering preclassification check to destroy the existing scan. nwokedi@chromium.org chrome-counter-abuse-alerts@Google.com Logs what the IntelligentScanVerdict from ClientPhishingResponse was. This is only logged in Client Side Phishing Protection after the server response comes back and the field IntelligentScanVerdict is populated. jdh@chromium.org nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the reason why visual phishing classification returned early on iOS before ML scoring could complete, such as a missing scorer or snapshot generation failure. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether or not the page of supported mime type is an error document. This is recorded on every supported mime type pages in the PreClassificationCheck. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether the intelligent scan is available at the time of inquiry. This is logged in the service class at the start of the inquiry function call requested by {IntelligentScanRequestType}. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records whether the on device model is available at the time of session creation. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether the intelligent_scan field under LlamaForcedTriggerInfo is set to true or false. This is recorded when the verdict contains the LlamaForcedTriggerInfo. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the field llama_trigger_rule_id under LlamaTriggerRuleInfo. This is recorded for every LlamaTriggerRuleInfo under LlamaForcedTriggerInfo. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the size of the LlamaTriggerRuleInfo repeated fields. This is recorded when the verdict contains the LlamaForcedTriggerInfo. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether the local phishing detection model detects phishing for a specific {RequestType}. This histogram is logged each time the model is run (roughly on every navigation to a non-allowlisted site) This metric is used to populate a dashboard on go/crsb-site. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the CSD allowlist match result for a specific {RequestType}. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Record the match result of allowlists for specific {RequestType}. This is logged during the client side detection host PreClassificationCheck. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether a dynamic update is successful or not. This is logged when a new model is pushed (rare), or on each startup. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether a dynamic update of the image embedding model is successful or not. This is logged when a new image embedding model is fetched, or on each startup. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the the duration of a network request due to a client side phishing ping. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the net error or HTTP response code after sending a network request due to a client side phishing ping. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether the on-device model update is successful or not. This happens for Enhanced Safe Browsing enabled users. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the OptimizationGuideOnDeviceModelEligibilityReason when the model is not available at the time of download failure. This happens in the intelligent scan delegate class when the model eligibility reason does not match what we're looking for, thus the download fails. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the OptimizationGuideOnDeviceModelEligibilityReason when the model is not available at the time of inquiry. This happens in the client side detection service class when we inquire the model, but the model is not available. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the duration it took to execute the on-device model. This happens for Enhanced Safe Browsing enabled users when a keyboard lock occurs on the page they are on and the on-device model session creation is available. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether the on-device model execution is successful or not. This happens for Enhanced Safe Browsing enabled users who called to execute the on-device model when inquiring for scam detection. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the duration it took to fetch the on-device model after subscription. This is only logged if the update is successful. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether the on-device model execution result's response parsing is successful or not. This is only logged when the model execution is successful. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether or not the on-device model session was alive when the delegate is shutting down. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether the on-device model session creation is successful. This is logged when the host class calls the inquire on device model function in the service class and the on device model is available. This is expected to be overwhelmingly "success" since we have passed the on device model availability check by the time this is logged. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the duration it took to create an on-device model session. This happens for Enhanced Safe Browsing enabled users when a keyboard lock occurs on the page they are on and the on-device model session creation is available. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records the duration it took to run the StartModelDownload function in the intelligent scan delegate. Note that this measures the function running time, not the actual model download time (which is measured by SBClientPhishing.OnDeviceModelFetchTime). nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the state of the client side detection host callback object when we respond after the on device model inquiry execution callback completed and the parsing was successful. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records the reason why the on-device model is unavailable at the time of inquiry on Android. This is logged when IsIntelligentScanAvailable is called with log_failed_eligibility_reason=true and the model is not available. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the duration of model fetch time after the OptimizationGuide Model Observer for CSD-Phishing is registered on user session start. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether the stripped GURLs with empty path on browser and renderer loaded URL would have matched when the reference stripped URLs have not. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the duration it takes for client side phishing detection requested by {RequestType} to finish. This occurs at the client side detection PreClassificationCheck. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com The result enum received from the renderer-side classifier. This is recorded after every renderer classification (roughly every page load not on the allowlist). {RequestType} indicates which process requested the renderer-side classification. This metric is used to populate a dashboard on go/crsb-site. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the duration it takes for image embedding requested by {RequestType} to finish. This occurs for ESB enabled users only after the image embedding request has been made. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com The result enum received from renderer-side image embedder when requested by {RequestType}. This is recorded after every renderer image embedding which is expected to happen when the user is subscribed to ESB, has an image embedding model loaded, and the page is deemed phishy by the phishing classification process. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the cancelling actor of type ClientSideDetectionType when there is a current preclassification check being done by {RequestType}. This is logged whenever an active preclassification check is cancelled. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the result of phishing pre-classification checks for a given {RequestType}. This is recorded on every classification attempt for roughly every {RequestType} that comes in. Starting in M95, the PRIVATE_IP bucket was split to include LOCAL_RESOURCE and CHROME_UI_PAGE. Starting in M137, the children histograms start to log the CLASSIFY bucket. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Logs whether the redirect chain from the classification URL contains URLs that has a ForcedTriggerInfo from RTLookupResponse. This is only logged when the URL that the user is on does not have a ForcedTriggerInfo and the URL contains a redirect chain other than itself. This is only for ESB users. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Logs whether the redirect chain from the classification URL contains URLs that has a force request from RTLookupResponse. This is only logged when the URL that the user is on does not have a force request and the URL contains a redirect chain other than itself. This is only for ESB users. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the request type that would have passed preclassification if not blocked by the report limit. This is logged whenever too many reports preclassification check value is logged. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records if the RTUrlLookupResponse contains FORCE_REQUEST in client_side_detection_type, which forces the Client Side Detection - Phishing ping to be sent to the server if the user also has Enhanced Safe Browsing or Safe Browsing Extended Reporting opt-in. This histogram is logged once on every successful classification. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records if the RTUrlLookupResponse that contained FORCE_REQUEST in client_side_detection_type contains the ForcedLlamaTriggerInfo. This histogram is logged if the RTUrlLookupResponse contains FORCE_REQUEST in client_side_detection_type and in the client side detection host class when the FORCE_REQUEST is fetched from the verdict cache manager. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether the URL at the primary main frame remained the same when PrimaryPageChanged occurred on a non-refresh navigation. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether the server-side phishing detection model detects phishing for a specific {RequestType}. This histogram is logged each time we contact Safe Browsing to evaluate whether a site is phishing. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records the duration it took to execute the server-side model. This happens for Enhanced Safe Browsing enabled users who called to execute the server-side model when inquiring for intelligent scan. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records the error returned in the server-side model execution response for intelligent scan. Recorded every time an error is encountered during server-side model execution. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records whether the server-side model execution is successful or not. This happens for Enhanced Safe Browsing enabled users who called to execute the server-side model when inquiring for intelligent scan. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records whether the user has hit the quota for intelligent scan when they try to start an intelligent scan. This is logged if the server-side model is available. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records whether the pref list for intelligent scan timestamps is empty when we try to remove a timestamp from it. This is logged when a scam warning is shown to the user, which refunds the quota. This should never happen if the pref is intact. bcl@google.com chrome-counter-abuse-alerts@google.com Records the number of server-side model queries made in the current 24-hour window for the user at the time a new server-side model lookup is initiated. The count includes the current lookup. It does not include refunded quota (i.e. from lookups which showed a warning). This is not recorded if the lookup is rejected due to hitting the quota limit. xinghuilu@chromium.org chrome-counter-abuse-alerts@google.com Records the number of server-side model queries made before a scam warning is shown to the user. This is logged when a scam warning is shown to the user, before the quota is refunded. This number can help inform how adjusting the quota size affects protection (i.e. how many warning we will lose if we reduce the quota). nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the single observer that was fired when there is a new page loading and the other observer CSD listens to has not been called. This is only called when one of the observers is fired and not the other. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com When CSD service hits the ping limit and we are adding another entry, record the time since last record. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records when a page load trigger CSD ping request is converted to a force request trigger. This is logged {LogTime}. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records the success of parsing the verdict string received from renderer classification. This metric is used to validate that the verdict can always be successfully parsed when the phishing detector result is SUCCESS nwokedi@chromium.org chrome-counter-abuse-alerts@google.com The width x height of the viewport encoded as an integer. The upper 16 bits is the width and the lower 16 bits is the height. Decode using bitwise operations in queries. It is recorded whenever CSD checks if visual features can be extracted. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com The height of the viewport. It is recorded whenever CSD checks if visual features can be extracted. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com The pixels per inch (PPI) of the display where the viewport is located. It is recorded whenever CSD checks if visual features can be extracted. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com The width of the viewport. It is recorded whenever CSD checks if visual features can be extracted. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Log the reason why CSD-Phishing pings do not contain visual features images. This is logged after deciding that the ping is eligible to be sent and the visual features extraction capability has been determined. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records whether visual features exist after image embedding. This is recorded after every image embedding. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records how long it took to extract visual features on the main thread. Work performed off-thread is not recorded in this histogram. This is recorded on every phishing classification or image embedding. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Logs the duration in PlaybackOnBackgroundThread when extracting visual features. This is recorded on every phishing classification or image embedding. This logs wall time while SBClientPhishing.VisualFeatureTime.BackgroundPlayback.ThreadDuration logs thread time. This metric is logged even for clients that do not support ThreadTicks or have high resolution clocks. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Logs the thread duration in PlaybackOnBackgroundThread when extracting visual features. This is recorded on every phishing classification or image embedding. This logs thread time while SBClientPhishing.VisualFeatureTime.BackgroundPlayback.Duration logs wall time. This metric is only recorded for clients with ThreadTicks support. Only recorded for clients with high resolution clocks. nwokedi@chromium.org chrome-counter-abuse-alerts@google.com Records how long it took to extract visual features. This logs the total duration, including the main thread time that is logged separately in SBClientPhishing.VisualFeatureTime, as well as the time taken on the background thread, up to but not including the execution of the done callback. This also includes queueing time for the posted task. This is recorded on every phishing classification or image embedding.